Workflows

[kbodwin]

I think the next short-term step is to let these models be part of a workflow with a recipe. Right now there's an error from add_model(): "Error: spec must be a model_spec."

I'm guessing that cluster_spec can't (and shouldn't?) be a subclass of model_spec, so that would mean that either add_model() needs a S3 method for cluster specs, or we need a add_celery_model()

[DavisVaughan]

Some thoughts:

• workflows has been defined in "stages" to sort of account for this possibility. There is a "fit stage" where you can swap out the thing used to fit. So I think we are going to have a fit-action-celery.R file dedicated to celery, like how there is a fit-action-model.R file dedicated to parsnip model specs. This allows celery to differ from parsnip when it matters, and we can easily account for those differences in workflows because there is a dedicated code path just for celery.
• control_workflow() currently has a control_parsnip argument. I think we could rename this to control to make it more generic, and allow a control_celery object through here too (it would become agnostic to the type of the actual control object and rely on the underlying fit() method to check the control object type)
• add_variables() requires the outcomes argument, but you can do outcomes = NULL to say "no outcome". I think that might be the best we can do there

[github-actions]

This issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with a reprex: https://reprex.tidyverse.org) and link to this issue.